Earthquake Damage Grades Prediction using An Ensemble Approach Integrating Advanced Machine and Deep Learning Models

Anurag Panda; Gaurav Kumar Yadav

REVIEW 4 major objections 6 minor 25 references

Earthquake Damage Grades Prediction using An Ensemble Approach Integrating Advanced Machine and Deep Learning Models

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims a stacking ensemble of standard classifiers predicts earthquake damage grades with 96% test accuracy on the Gorkha Nepal building dataset, outperforming ten other algorithms.

desk verdict A standard benchmark on a public dataset whose headline 96% accuracy is unverifiable because the evaluation protocol is missing and the references look unreliable. read the letter →

arxiv 2506.22129 v1 pith:6OCXF7QI submitted 2025-06-27 cs.LG

Anurag Panda , Gaurav Kumar Yadav This is my paper · ORCID

classification cs.LG

keywords earthquakedamagepredictionstackingensembleSMOTEclassimbalanceGorkhadatasetXGBoostmachinelearningdeep

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper tries to show that a stacking ensemble of ordinary machine-learning classifiers can predict the damage grade of buildings after an earthquake with 96% test accuracy, using only building attributes such as age, construction materials, and floor type. The test bed is the dataset compiled after the 2015 Gorkha, Nepal, earthquake. The authors argue that handling class imbalance with SMOTE, selecting the top 20 features, and combining several classifiers through a meta-learner gives a practical tool for post-disaster triage and relief prioritization. If the result holds, a cheap, fast model could flag the most damaged buildings before inspectors reach them.

What carries the argument

The load-bearing mechanism is the stacking ensemble: a set of base classifiers (Random Forest, Gradient Boosting, LightGBM, XGBoost, and others) each produce damage-grade predictions, and a meta-learner learns to combine those outputs into the final grade. Around it, the pipeline uses SMOTE (synthetic minority oversampling) plus RandomUnderSampler to rebalance the three damage classes, Isolation Forest to remove anomalous rows, and SelectKBest with the ANOVA F-test to keep the 20 most informative building features. The deep-learning alternatives are a feedforward neural network and a Kolmogorov-Arnold Network, which transforms each input dimension before combining them.

What would settle it

Apply the pipeline with a train/test split fixed before SMOTE, undersampling, and feature selection, and compare test accuracy against a run where these steps see the full dataset; a large drop in the untouched-holdout run would show the 96% figure is inflated.

Watch

Extended reading notes

Core claim

The paper reports that a Stacking Classifier, which feeds predictions from Random Forest, GBM, LightGBM, XGBoost, and other base models into a meta-learner, achieves 96% test accuracy and 99.9% training accuracy, the best among the ten algorithms compared. Random Forest, GBM, LightGBM, Voting, and Bagging also reach roughly 96%, while Logistic Regression and AdaBoost lag. Among deep models, a feedforward network reaches 92.4% test accuracy and a Kolmogorov-Arnold Network reaches 94.39% test accuracy. The authors attribute the result to balancing the training data with SMOTE and RandomUnderSampler, removing anomalies with Isolation Forest, selecting the top 20 features by ANOVA F-test, and tuning hyperparameters.

Load-bearing premise

The load-bearing premise is that SMOTE and undersampling were applied only to the training portion, so the reported 96% test accuracy comes from a holdout set the balancing steps never saw; the paper does not say where that split occurs.

Editorial extensions

If this is right

If the 96% test accuracy is genuine, post-earthquake teams could rank buildings by predicted damage grade from survey attributes alone, focusing field inspections on the most likely severe cases.
The same SMOTE-plus-ensemble pipeline is claimed to transfer to other natural calamities, giving a flexible modelling template for disaster response.
Feature selection points to building age, construction materials, foundation, roof, and ground-floor type as the attributes that matter most for seismic vulnerability.
The near-tie among stacking, Random Forest, GBM, and LightGBM suggests that a single strong tree-based model may capture most of the signal, with stacking adding a modest edge.
Deep learning models, while behind the best ensembles, still reach above 92% test accuracy, so the choice between ML and DL depends on interpretability and compute constraints.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A consequence the paper does not test: the reported 96% could be optimistic if resampling touched the test set, since the paper never states the split point; re-running with a strictly leakage-free split would settle it.
Because the dataset comes from one earthquake and one region, the pipeline's value for other seismic zones is an open question; a natural next experiment is applying it to damage data from a different earthquake.
The gap between 99.9% training accuracy and 96% test accuracy hints that the stacking model may be overfitting or memorizing, so a simpler model or stronger regularization might generalize just as well on new data.
One useful extension the paper leaves implicit: reporting per-class accuracy after balancing, since overall accuracy can hide poor performance on rare damage grades.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Thanks for the report. I read the paper. Bottom line: this is a routine benchmark on the Ghimire et al. Gorkha dataset, and I wouldn't trust the headline 96% test accuracy because the paper never says where the train/test split is relative to SMOTE and undersampling, and the tables don't label train vs test. The stress test is right.

What's actually here: the authors apply a standard pipeline—label encoding, anomaly removal, SMOTE + RandomUnderSampler, SelectKBest, then 10 classifiers and ensembles—to a published dataset. That's fine as a technical exercise. The positive side is they compare a reasonable set of models including KAN, which you don't see often in this literature, and they report per-class precision/recall/F1. The dataset is real and public, so someone could reproduce this if they had the code. That's about it.

The problems are load-bearing. The central claim—96% test accuracy for stacking—is not traceable. Section 4.1 lists SMOTE and undersampling before any mention of a holdout. If resampling happens before splitting, test points are contaminated by synthetic neighbors from the training set, and every number in the paper is inflated. The k-fold description in Section 2.6 is generic: no k, no nesting, no ordering. Tables 3–5 report the same 'accuracy' value for each algorithm across all three class tables, which makes no sense as per-class accuracy and suggests the numbers are copied from a single run, possibly training accuracy. The conclusion states 99.9% training accuracy for stacking, but that number doesn't appear in any table. No code, no baselines, no error bars. These aren't minor omissions; they make the headline result unverifiable.

I also can't overlook the references. Many entries have the same suspicious page pattern (149:0402308x), author names like 'Min Lu and et al.', and 'Wei Zhang and et al.' These read like placeholders or fabricated citations. One of the authors' own papers is cited for SMOTE, which is odd. A reviewer would need to verify each one.

Who is this for? Someone building a quick baseline for earthquake damage prediction might skim it, but they'd have to redo the whole pipeline themselves. As a research contribution, the paper is a parameter scan with an unclear evaluation. I'd desk reject it on the protocol gap alone; the reference list makes that decision easier. If the authors resubmit with a clear split-before-resampling statement, labeled tables, code, and a cleaned reference list, it could be a modest but citable benchmark. As is, I wouldn't send it to a serious referee.

Referee Report

4 major / 6 minor

Summary. The manuscript proposes a pipeline for classifying building damage grades into three classes using the 2015 Gorkha earthquake dataset, combining SMOTE, RandomUnderSampler, Isolation Forest, and SelectKBest feature selection with base classifiers (Logistic Regression, Decision Tree, Random Forest, GBM, AdaBoost, LightGBM, XGBoost), ensemble methods (Voting, Bagging, Stacking), and deep models (FFN and KAN). The central claim, stated in Section 6, is that a Stacking Ensemble achieves 96% test accuracy and 99.9% training accuracy, outperforming ten other algorithms. The paper provides mathematical formulations for SMOTE, loss functions, ensemble methods, and evaluation metrics, but the experimental evaluation is critically underspecified.

Significance. If the 96% test accuracy claim were validated, it would suggest that a stacking ensemble of standard classifiers is a highly accurate tool for post-earthquake building damage triage, which could be practically useful for disaster response. The paper uses a publicly available real-world dataset and explores a broad set of models, including a comparison between classical ML and newer KAN architectures. However, the evaluation protocol is not described with sufficient detail to verify the central quantitative result, and no code, error bars, or statistical comparisons are provided; as it stands, the significance of the claimed result cannot be assessed.

major comments (4)

[Section 4.1] The preprocessing pipeline described in Section 4.1 applies LabelEncoder, Isolation Forest, SMOTE with RandomUnderSampler, and SelectKBest before any mention of a train/test split, while Section 2.6 defines only generic k-fold cross-validation with no specified k and no nested procedure. This creates a leakage risk: if SMOTE or undersampling is applied before splitting, synthetic samples derived from all original data can appear in both training and test partitions, inflating every reported metric. The paper never states where the split occurs, so the claimed 96% test accuracy in Section 6 cannot be identified as coming from an untouched holdout set. This is a load-bearing gap that directly undermines the central claim.
[Tables 3-5] Tables 3, 4, and 5 are labeled as per-class performance for Class 0, Class 1, and Class 2, but the 'Accuracy' column contains identical values for each algorithm across all three tables (e.g., Stacking Classifier is 0.96 in every table). Accuracy is a global metric, so labeling these tables as per-class accuracy is incoherent, and no header or text states whether the numbers are train, validation, or test results. In addition, the conclusion's 99.9% training accuracy appears nowhere in the results tables, so the Section 6 claim cannot be traced to any reported result.
[Section 5] The comparison across the ten algorithms in Section 5 is presented without error bars, confidence intervals, repeated runs over random seeds, or any statistical significance test. In Tables 3-5, most strong models are clustered at accuracy 0.95-0.96, and the differences between the Stacking Classifier and Random Forest or LightGBM are within one percentage point, which is likely within noise. The assertion in Section 6 that the Stacking Ensemble 'stands out' among ten algorithms is therefore unsupported by the evidence presented.
[Section 5.2] The FFN and KAN results in Sections 5.2.1 and 5.2.2 report train and test accuracies (94%/92.4% for FFN, 96.44%/94.39% for KAN) but give no description of the train/test split, whether SMOTE was applied before or after splitting, or the hyperparameter search protocol for these deep models. The KAN's 96.44% training accuracy is also quoted in the conclusion as an example of 'strong generalization,' but without a clear evaluation protocol it is impossible to determine whether the reported test accuracies are honest holdout results.

minor comments (6)

[Section 2.2] The text contains a typo in 'ANOV A F-test'; it should read 'ANOVA F-test'.
[Section 2.6] The number of folds k in the k-fold cross-validation is never specified, and no nested cross-validation procedure is described for cases where feature selection or hyperparameter tuning is performed.
[Tables 3-5] The table headers contain spacing errors such as 'W eighted Avg', and the 'Macro Avg' and 'Weighted Avg' columns are not defined in Section 2.4.
[Section 1] The paper's outline in Section 1 says that Section 2 reviews related works, Section 3 presents methodology, and Section 4 presents results, but the actual sections are shifted: Section 2 is mathematical formulation, Section 3 is related works, Section 4 is methodology, and Section 5 is results; the outline should be corrected.
[Table 1] Table 1 reports a count of 100 for every numerical feature, which is inconsistent with the categorical frequency counts in Table 2 (e.g., 254,477 occurrences) and with the presumably much larger dataset size; the sampling or subset used for the statistical summary should be clarified.
[References] Several references are duplicated or inconsistently formatted, including 'Lu and et al.' versus 'Min Lu and et al.', and 'Klusowski and Tian' appears as both 2023 and 2024 entries; also, the paper provides no link to code or data availability.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper reports empirical classifier benchmarks; the unclear train/test split is a soundness concern, not a circular derivation.

full rationale

This paper does not claim to derive damage-grade predictions from first principles; it trains standard classifiers and ensembles on a public dataset and reports empirical metrics. The mathematical formulation (Sections 2.1-2.6) defines SMOTE, SelectKBest, loss functions, ensembling, and evaluation metrics, but no result is defined in terms of another result, and no fitted parameter is renamed as a prediction. The only self-citation, Yadav et al. [2023] for SMOTE in Section 4.1, is not load-bearing: SMOTE is a standard algorithm and is independently specified by the interpolation formula in Section 2.1. The central stacking-ensemble claim (Section 6) is presented as an observed test accuracy, not as a quantity constructed from its own inputs. The manuscript's failure to state whether SMOTE and undersampling were applied before or after the train/test split is a potentially serious evaluation-protocol gap that could inflate all reported numbers, but that is a soundness and validity concern, not a circularity: the reported accuracy is not equivalent to its own inputs by construction. No uniqueness theorem, ansatz smuggled in via citation, or renamed known result appears in the paper.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The paper introduces no new entities, particles, or forces. Its empirical claims rest on standard ML tools, tuned hyperparameters, and several dataset assumptions. The free parameters are typical ML hyperparameters, but they are not documented well enough to reproduce the exact numbers.

free parameters (5)

Number of selected features (k) = 20
SelectKBest with ANOVA F-test selected the top 20 features; this choice affects all model inputs and is not justified by any analysis.
FFN L2 regularization lambda = 4.918e-3
Reported in Section 5.2.1 as optimal after tuning; a fitted value that influences the reported accuracy.
FFN learning rate = 3.885e-4
Reported as optimal in Section 5.2.1.
FFN dropout rate = 0.1033
Reported as optimal in Section 5.2.1.
KAN learning rate = 0.01
Reported in Section 5.2.2 as the chosen learning rate for the KAN model.

assumptions (4)

domain assumption SMOTE-generated synthetic samples are valid representatives of minority classes and improve generalization to real test data.
Section 2.1 and Section 4.1 rely on this interpolation assumption; if synthetic samples do not lie on the true decision boundary, reported accuracy is an artifact.
domain assumption The Gorkha building dataset's labels and features are accurate and representative of regional seismic vulnerability.
Section 4.2 uses this dataset without external validation of labels; the entire study depends on it.
domain assumption Isolation Forest correctly identifies anomalies without discarding informative hard cases.
Section 4.1 applies Isolation Forest before training but provides no analysis of what was removed or how that affects results.
standard math Standard probability and cross-entropy loss definitions apply to the classification problem.
Section 2 uses standard definitions for loss, SMOTE interpolation, and performance metrics.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Earthquake Damage Grades Prediction using An Ensemble Approach Integrating Advanced Machine and Deep Learning Models." pith.science (2026). https://pith.science/paper/6OCXF7QI

@misc{pith2026250622129,
  author       = {Pith},
  title        = {Pith review of: Earthquake Damage Grades Prediction using An Ensemble Approach Integrating Advanced Machine and Deep Learning Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6OCXF7QI}},
  note         = {Machine review of arXiv:2506.22129}
}

read the original abstract

In the aftermath of major earthquakes, evaluating structural and infrastructural damage is vital for coordinating post-disaster response efforts. This includes assessing damage's extent and spatial distribution to prioritize rescue operations and resource allocation. Accurately estimating damage grades to buildings post-earthquake is paramount for effective response and recovery, given the significant impact on lives and properties, underscoring the urgency of streamlining relief fund allocation processes. Previous studies have shown the effectiveness of multi-class classification, especially XGBoost, along with other machine learning models and ensembling methods, incorporating regularization to address class imbalance. One consequence of class imbalance is that it may give rise to skewed models that undervalue minority classes and give preference to the majority class. This research deals with the problem of class imbalance with the help of the synthetic minority oversampling technique (SMOTE). We delve into multiple multi-class classification machine learning, deep learning models, and ensembling methods to forecast structural damage grades. The study elucidates performance determinants through comprehensive feature manipulation experiments and diverse training approaches. It identifies key factors contributing to seismic vulnerability while evaluating model performance using techniques like the confusion matrix further to enhance understanding of the effectiveness of earthquake damage prediction.

Figures

Figures reproduced from arXiv: 2506.22129 by the authors.

**Figure 1.** Model Architecture 4.2 The Data The earthquake damage prediction dataset Ghimire et al. [2022] contains various features describing the structural and situational characteristics of buildings, aimed at understanding the factors influencing earthquake-induced damage. This analysis focuses on key numerical and categorical features, providing insights into their distributions and potential correlations with the targe… view at source ↗

**Figure 2.** Heatmap Categorical Features The categorical and binary features include various attributes such as construction materials, structural design, and legal ownership status. There are 10 unique plan configurations, with the most common appearing 86,884 times. Binary indicators for superstructure materials show low usage of mud mortar stone, mud mortar brick, cement mortar brick, timber, or bamboo, mostly indicated by 0… view at source ↗

**Figure 3.** Training and Test Accuracy Statistics 12 [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Training and Test Accuracy Statistics 6 Conclusions This study enhances earthquake damage prediction by integrating machine learning and deep learning techniques. Through ensemble methods and rigorous feature engineering, robust models accurately forecast damage grade…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

25 extracted references · 25 canonical work pages

[1]

T. Chen, J. Li, and C. Wang. Deep reinforcement learning for real-time earthquake damage assessment. Journal of Structural Engineering, 149: 0 04023087, 2023

work page 2023
[2]

W. Gao, X. Liu, and Z. Wang. Neural network-based approach for earthquake damage prediction. Journal of Geotechnical and Geoenvironmental Engineering, 149: 0 04023093, 2023

work page 2023
[3]

Garcia, J

E. Garcia, J. Martinez, and M. Lopez. A comparative study of machine learning models for earthquake damage prediction. Natural Hazards, 88: 0 1245--1260, 2023

work page 2023
[4]

Testing machine learning models for seismic damage prediction at a regional scale using building-damage dataset compiled after the 2015 gorkha nepal earthquake

Subash Ghimire, Philippe Gu \'e guen, Sophie Giffard-Roisin, and Danijel Schorlemmer. Testing machine learning models for seismic damage prediction at a regional scale using building-damage dataset compiled after the 2015 gorkha nepal earthquake. Earthquake Spectra, 38 0 (4): 0 2970--2993, 2022

work page 2015
[5]

Huang, L

Q. Huang, L. Zhang, and Y. Chen. Ensemble learning for seismic damage prediction of buildings. Earthquake Engineering and Engineering Vibration, 22: 0 625--641, 2023

work page 2023
[6]

J. Kim, S. Lee, and H. Park. Application of artificial intelligence techniques in earthquake damage prediction. Journal of Structural Engineering, 149: 0 04023092, 2023

work page 2023
[7]

Scalable decision trees for earthquake damage assessment

Jason Klusowski and Yuan Tian. Scalable decision trees for earthquake damage assessment. Journal of Machine Learning Research, 24 0 (1): 0 1--20, 2023

work page 2023
[8]

Large scale prediction with decision trees

Jason M Klusowski and Peter M Tian. Large scale prediction with decision trees. Journal of the American Statistical Association, 119 0 (545): 0 525--537, 2024

work page 2024

Show all 25 references

[9]

C. Li, G. Zhang, and S. Wu. Probabilistic modeling of earthquake-induced building damage: A review. Journal of Performance of Constructed Facilities, 149: 0 04023091, 2023 a

2023
[10]

Y. Li, Q. Wu, and Y. Zhang. Performance comparison of machine learning models for earthquake damage prediction. Natural Hazards, 89: 0 235--250, 2023 b

2023
[11]

F. Liu, H. Zhang, and X. Wu. Application of machine learning techniques in earthquake damage assessment: A review. Journal of Geotechnical and Geoenvironmental Engineering, 149: 0 04023088, 2023 a

2023
[12]

Q. Liu, H. Wang, and Y. Zhang. Comparison of machine learning models for seismic damage prediction. Natural Hazards Review, 24: 0 247--262, 2023 b

2023
[13]

H. Lu, Y. Wang, and Z. Li. Earthquake damage prediction using xgboost. Journal of Seismology, 27: 0 123--136, 2023

2023
[14]

Xgboost: A scalable machine learning system for tree boosting

Min Lu and et al. Xgboost: A scalable machine learning system for tree boosting. IEEE Transactions on Machine Learning, 15 0 (1): 0 1--12, 2023

2023
[15]

J. Park, S. Kim, and D. Lee. Data-driven approaches for seismic vulnerability assessment of buildings. Journal of Earthquake Engineering, 149: 0 04023089, 2023

2023
[16]

Singh and R

A. Singh and R. Gupta. Ensemble methods for earthquake damage prediction. Natural Hazards Review, 24: 0 215--230, 2023

2023
[17]

X. Wang, P. Li, and P. Jiang. Resilience assessment of urban infrastructure systems under earthquake hazard. Journal of Urban Planning and Development, 149: 0 04023086, 2023 a

2023
[18]

Y. Wang, H. Chen, and L. Li. Predictive modeling of earthquake damage using bayesian networks. Journal of Structural Engineering, 149: 0 04023090, 2023 b

2023
[19]

Y. Wang, S. Zhang, and H. Liu. Machine learning approaches for post-earthquake damage assessment. Journal of Structural Engineering, 149: 0 04023096, 2023 c

2023
[20]

Effective ml-based quality of life prediction approach for dependent people in guardianship entities

Gaurav Kumar Yadav, Benigno Moreno Vidales, Hatem A Rashwan, Joan Oliver, Domenec Puig, GC Nandi, and Mohamed Abdel-Nasser. Effective ml-based quality of life prediction approach for dependent people in guardianship entities. Alexandria Engineering Journal, 65: 0 909--919, 2023

2023
[21]

J. Yang, X. Chen, and L. Zhu. Earthquake damage prediction using machine learning techniques: A comprehensive review. Journal of Earthquake Engineering, 149: 0 04023095, 2023

2023
[22]

Q. Yin, S. Zhang, and H. Liu. Predicting earthquake damage using support vector machines. Geophysical Research Letters, 50: 0 2876--2883, 2023

2023
[23]

A cnn-based approach for earthquake damage prediction

Wei Zhang and et al. A cnn-based approach for earthquake damage prediction. Seismic Engineering Journal, 18 0 (4): 0 45--60, 2023

2023
[24]

J. Zhao, Y. Liu, and Q. Zhang. Machine learning for post-earthquake damage estimation: A case study. Journal of Disaster Research, 18: 0 520--535, 2023

2023
[25]

H. Zhou, X. Wang, and T. Zhang. Machine learning models for earthquake damage assessment: A comparative study. Natural Hazards Review, 24: 0 231--246, 2023

2023

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

T. Chen, J. Li, and C. Wang. Deep reinforcement learning for real-time earthquake damage assessment. Journal of Structural Engineering, 149: 0 04023087, 2023

work page 2023

[2] [2]

W. Gao, X. Liu, and Z. Wang. Neural network-based approach for earthquake damage prediction. Journal of Geotechnical and Geoenvironmental Engineering, 149: 0 04023093, 2023

work page 2023

[3] [3]

Garcia, J

E. Garcia, J. Martinez, and M. Lopez. A comparative study of machine learning models for earthquake damage prediction. Natural Hazards, 88: 0 1245--1260, 2023

work page 2023

[4] [4]

Testing machine learning models for seismic damage prediction at a regional scale using building-damage dataset compiled after the 2015 gorkha nepal earthquake

Subash Ghimire, Philippe Gu \'e guen, Sophie Giffard-Roisin, and Danijel Schorlemmer. Testing machine learning models for seismic damage prediction at a regional scale using building-damage dataset compiled after the 2015 gorkha nepal earthquake. Earthquake Spectra, 38 0 (4): 0 2970--2993, 2022

work page 2015

[5] [5]

Huang, L

Q. Huang, L. Zhang, and Y. Chen. Ensemble learning for seismic damage prediction of buildings. Earthquake Engineering and Engineering Vibration, 22: 0 625--641, 2023

work page 2023

[6] [6]

J. Kim, S. Lee, and H. Park. Application of artificial intelligence techniques in earthquake damage prediction. Journal of Structural Engineering, 149: 0 04023092, 2023

work page 2023

[7] [7]

Scalable decision trees for earthquake damage assessment

Jason Klusowski and Yuan Tian. Scalable decision trees for earthquake damage assessment. Journal of Machine Learning Research, 24 0 (1): 0 1--20, 2023

work page 2023

[8] [8]

Large scale prediction with decision trees

Jason M Klusowski and Peter M Tian. Large scale prediction with decision trees. Journal of the American Statistical Association, 119 0 (545): 0 525--537, 2024

work page 2024